Research Review: Does Trend Following Work on Stocks?

In November 2005 Cole Wilcox and Eric Crittenden of Blackstar Funds LLC* (now Longboard Asset Management) published a research report analyzing the effectiveness in using a Trend Following trading strategy in the US equity markets.  Both fund managers were using Trend Following successfully in the futures markets for many years.  Their success with Trend Following, as well as their peer's results in similar markets, piqued their curiosity and led them to conduct this research. 

The strategy tested is a long-only Trend Following program. Trend Following uses absolute price change to delineate strength or weakness in a particular security. In this case, the researchers added long exposure on positive absolute price changes that resulted in an all-time high on a one week closing basis.

Before actual testing began, Wilcox and Crittenden made sure to address any data issues. For example, given the expansive time horizon for testing, the authors account for security-specific corporate actions (e.g. stocks splits, mergers and acquisitions), delisted companies were included to account for survivorship bias**, liquidity filters were applied to include only stocks with enough daily liquidity for institutional investors, and realistic transaction costs (i.e. slippage and commissions) were also accounted for during the study.

The methods and results are as follows:

Entry and Exit:

Initial long entries are taken when a stock reaches an all-time highest close on a weekly basis.  If today’s close is greater than or equal to the highest close during the stock’s entire history then buy tomorrow on the open.  Additional longs were added when subsequent weeks closed at new all-time highs. 

Blackstar Funds employed an Average True Range (ATR) for its trailing stops. This form of position level risk management is universally applicable and commonly used to minimize position volatility and tailor risk management protocol for Trend Following systems.  When trading, it is important to not risk too much on any one position. Volatility adjusting the position size with an ATR is an easy, yet effective way to keep your financial and emotional capital unharmed.

Average True Range is a derivative of the True Range indicator, which measures the daily movement by calculating the greater absolute values of:

·         Today’s high minus today’s low

·         Today’s high minus yesterday’s close

·         Yesterday’s close minus today’s low

A simple arithmetic average is then taken of the greater of these values over a specified period (e.g.  10 week period). A stop loss level is established through multiplying this ATR value by a constant (K)  which is usually a value of 1, 2, or 3, depending on the risk tolerance of the investor and time horizon for the trading system, and then subtracting the product (i.e. K x ATR) from the previous day’s high or entry price (for long positions).

When a stop loss level was triggered, the researchers opted to exit a stock on the open the day after the exit level was breached. 

Expectancy Study:

When reviewing testing results it is best to take the average of all samples (Trading Systems and Methods, Kaufman). The average trade results in this study produced a distribution exhibiting positive skew and leptokurtosis (i.e. a fat right tail). These results are not surprising given the premise behind Trend Following is taking small losses (i.e. conservation of capital) while letting winners run (i.e. capturing large outlier moves). The testing consisted of 24,000+ trades during a 22 year test period using 0.5% as the transaction cost per round-turn.  Specifically, the results are as follows:

Average Trade: +15.2% (i.e. positive expectancy)

Average Days in Average Trade: 305

Average Winning Trade: +51.2%

Average Days in Winning Trade: 441 (i.e. “long-term” capital gains)

Average Losing Trade: -20.0%

Average Days in Losing Trade: 175

Winning Percentage: 49.3% (Exceptionally high for a Trend Following system)

The ratio of average winning trade to average losing trade was 2.56:1.00. Most trading systems require a minimum of 2:1, so this strategy exceeds that criterion. Moreover, it can be seen that the system yields a positive mathematical expectancy of +15.2%. In other words, when analyzing the trading results for a system it should show a positive return in order to actually generate a return on investment.

While an ATR of 10 was used for the test results above, the range spanned from 8 to 12.  Lower ATR levels (tighter stop loss levels) resulted in slightly lower winning percentages and slightly higher win/loss ratios.  The converse was true of higher ATR levels (looser stops). The ideal ATR range will depend on the individual investor’s risk preference and a selection of the most robust parameters given different stop loss volatility ranges.

While there are inherent limitations with the system tested, the Blackstar study provides an excellent example for the process behind system design and testing. In other words, all elements of system design were not addressed but the results help prove the effectiveness of using a Trend Following strategy with US equities. It would behoove traders read further into the study and to use this research process as a template in their next system design and testing procedure.

As always, please feel free to contact me with any comments or questions. Thanks for reading.

JD

*Blackstar Funds LLC was initially backed by Tom Basso of TrendStat Capital Management. Tom was featured in The New Market Wizards by Jack Schwager, which profiled some of the most consistently successful traders up until 1992. While Blackstar was in existence, Tom acted as an advisor to Cole and Eric.

**Survivorship bias impacts indices where constituents that have been delisted due to bankruptcy or poor performance are no longer represented in a sample. This results in an overestimation of past performance.